An improved statistical model for predicting the deuterium ingress in zirconium alloy pressure tubes

In the CANDU pressurized heavy water reactor (PHWR), the nuclear fuel is contained in hundreds of Zr–2.5 Nb alloy pressure tubes. The corrosion of zirconium alloy produces deuterium that is absorbed by the body of the pressure tube. The presence of this deuterium causes hydrogen embrittlement of zirconium alloy with an adverse effect on the integrity of the pressure tube. An accurate prediction of deuterium accumulation over time is an important step for ensuring the fitness-for-service of pressure tubes. Deuterium ingress data collected from in-service inspection of pressure tubes exhibit heteroscedasticity, i.e., the variance of deuterium concentration is dependent on operating time (or exposure) and temperature. The currently used model by the nuclear industry involves a logarithmic regression of deuterium content over time and temperature. Since this approach does not deal with heteroscedasticity precisely, it results in a conservative prediction of the deuterium ingress. The paper presents a new approach for predicting deuterium ingress based on a weighted least-squares (WLS) regression that overcomes the limitations of the existing model, and it provides realistic prediction bounds of deuterium ingress.